# How to use latin hypercube sampling? Neural network optimization help?

9 views (last 30 days)
Samy Akrouh on 28 Dec 2018
Edited: Samy Akrouh on 2 Jan 2019
I have been running simulations for the past few days. It is about a drill simulation, and by changing speed and feed I calculated the material removal rate (MRR) and got the nodal temperatures (NT). My aim is to make a multiobjective optimization where maximizing MRR and minimizing NT depending on the feed and speed. How would i do this in MATLAB? I attached the data if that helps understanding my question.

Naman Chaturvedi on 2 Jan 2019
Since your simulation is quite time consuming and you are considering sampling methods as well, I would suggest an approach like Surrogate Modelling for your work.
You can also find some third party toolboxes/codes related to multi-objective surrogate model optimization methods using MATLAB for a use case similar to yours.
Hope this helps!
Samy Akrouh on 2 Jan 2019
Thank you very much Naman! It seems to be exactly what I am looking for.
Now, my input data are 3 variables and 2 objectives and the data i got is already numerical:
Maximixe F(X)={ f1(x)=x4, -f2(x)=x5} ;; [2000<x4x1000, 15000<x5<85000]
Subject to
-2500 <g1(x)=x1< -1500
-0.6<g2(x)=x2< -0.35
-7 <g3(x)=x3< -3
Being
x1=Speed
x2=Feed
x3=Depth of cut
x4=MRR
x5=NT
Each of them a vector of 25 values (in the attached file).
How would I adapt the datainput_mop2.m for this?
Thank you in advance and sorry for this probably stupid question =D